Proactive Virtual Resource Management In Cloud Based On Resource Utilization

With the rapid development of cloud computing,IT giants have benefited a lot from providing cloud services.At the same time,more and more application providers are transferring their applications to the cloud in order to be free from the burden of system administration.Meanwhile,the world is full of information explosion,which makes some applications become hot in a short period of time unexpectedly.Thus these applications in cloud may encounter sudden traffic increment or decrement.Usually,developers use the alarm service or the auto scaling service provided by the cloud provider to tackle sudden traffic change.But these are reactive methods which have time latency and usually they only consider horizontal resizing of scaling group.In this paper,we propose a proactive framework based on the prediction of resource amount to cope with sudden traffic change.Compared with post-action methods and existing proactive methods,we have lower time latency and we consider not only horizontal resizing but also vertical resizing of scaling group which makes our model quicker and much more cost-effective.Firstly,we present a resource amount prediction algorithm based on the combination of Gompertz curve fitting and moving average model.When the real resource amount is steady or dropping quickly,we use moving average model to do prediction.When the real resource amount is rising quickly,we use Gompertz curve fitting to do prediction and the parameters in Gompertz curve is computed by using non-linear least square method.Compared with existing prediction methods,our algorithm ensures in most cases,the predicted resource amount is slightly higher than the real need,ensures the performance of cloud applications when the traffic is rising quickly and also ensures cost-effective targets when the traffic is dropping down quickly.Secondly,we present resource provisioning algorithm to correct the deviation of predicted value.Resource provisioning contains resource bottleneck check and resource amount computing.Resource bottleneck check is used to check the existing resource bottleneck and resource amount computing is responsible for correcting the predicted value according to the bottleneck check results.Then,we present resource placement and recycling which combines vertical resizing and horizontal resizing of VM.Vertical resizing contains vertical scaling up and vertical scaling down.Horizontal resizing contains horizontal scaling up and horizontal scaling down.When the resource is to be scaling up,we first consider vertical scaling up because vertical scaling up is fast to deploy and configure.When the resource is to be scaling down,we first consider horizontal scaling down because horizontal scaling down may reduce the number of VM or even PM.Meanwhile,when vertical resizing is considered,we present resource class table to resolve the consolidation of different VM in the same PM.When horizontal resizing is considered,we present layered delegation model to find a proper PM in the cloud.Finally,we design simulation experiments.By comparing our work with other work,the results show that under sudden traffic changes,our model can predict sudden traffic change in advance,can add or recycle resources in time.Our model ensures the performance targets of applications in cloud and saves the use of resources for the targets of cost-effective.